real RAG answers on POST /api/ai/ask, plus SSE streaming and answer cache

When RAG is configured the endpoint retrieves lesson-scoped chunks and
generates a grounded answer; otherwise the previous deterministic stub
runs unchanged, so keyless runs and existing tests are unaffected.
Quota / idempotency / user-row lock / clean-503 behaviour is preserved on
every path.

models.py
  - Conversation.subject (CharField) and Conversation.lesson_number
    (IntegerField): retrieval metadata sent on the first ask and reused
    for the rest of the conversation.
migrations/0004_conversation_lesson_number_conversation_subject.py
  - migration for the two new fields.
serializers.py
  - AskRequestSerializer: optional subject / grade / lessonNumber to pin
    the Qdrant collection and lesson/chapter.
  - AskResponseSerializer: usedChunks (retrieved [{title, summary}]) and
    retrievalDebug (collection / filterKey / chapter-or-lesson / counts,
    emitted only under DEBUG).
views.py
  - generate_tutor_answer(): unchanged signature; RAG-configured branch
    calls rag.services generation, stub branch otherwise.
  - AskAIView.post(): resolve subject/grade/lessonNumber (falling back to
    the Conversation), check the answer cache via resolve_target(), run
    retrieve_context + generate_answer on a miss and populate the cache,
    return usedChunks (+ retrievalDebug under DEBUG). Failures still
    release the reserved quota slot and surface as 503.
  - AskAIView._streaming_response(): opt-in "stream": true returns a
    text/event-stream of chunk / done / error events; same retrieval,
    phase-3 persist, idempotency and quota accounting as the JSON path,
    and it consults the answer cache first. Non-stream JSON contract is
    untouched.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

ai_tutor/serializers.py

@@ -7,16 +7,37 @@ class AskRequestSerializer(serializers.Serializer):
     chapterId = serializers.CharField(required=False, allow_null=True, allow_blank=True)
     gradeId = serializers.IntegerField(required=False, allow_null=True)
     questionId = serializers.CharField(required=False, allow_null=True, allow_blank=True)
+    # metadataِ retrieval برای لایه‌ی RAG (اپ rag): برای زدن به دیتای واقعیِ
+    # کتاب‌ها در Qdrant باید subject (مثلاً «علوم تجربی») + gradeId + lessonNumber
+    # همگی مطابقِ دادهٔ موجود ارسال بشن؛ اگه نباشن، پاسخ به‌صورتِ تدریسِ عمومی
+    # (بدونِ استناد به کتاب) تولید می‌شه. در ادامه‌ی همون مکالمه لازم نیست
+    # دوباره فرستاده بشن (روی Conversation ذخیره می‌شن).
+    subject = serializers.CharField(required=False, allow_null=True, allow_blank=True)
+    # جزءِ «پایه» در نامِ collectionِ Qdrant (مثلاً «ششم»). اگه بیاد مستقیم
+    # استفاده می‌شه؛ اگه نه، از روی gradeId و content.Grade.title / نگاشتِ
+    # RAG_GRADE_TITLE_OVERRIDES ساخته می‌شه.
+    grade = serializers.CharField(required=False, allow_null=True, allow_blank=True)
+    lessonNumber = serializers.IntegerField(required=False, allow_null=True)
     userQuestion = serializers.CharField()
     stream = serializers.BooleanField(required=False, default=False)
 
 
+class UsedChunkSerializer(serializers.Serializer):
+    title = serializers.CharField(allow_blank=True)
+    summary = serializers.CharField()
+
+
 class AskResponseSerializer(serializers.Serializer):
     conversationId = serializers.CharField()
     messageId = serializers.CharField()
     answer = serializers.CharField()
     suggestedFollowUps = serializers.ListField(child=serializers.CharField())
     remainingQuota = serializers.IntegerField()
+    # تکه‌های درسیِ بازیابی‌شده از Qdrant که پاسخ بر پایه‌شون ساخته شده؛
+    # خالی یعنی retrieval انجام نشده (نبودِ subject/lessonNumber یا خاموش‌بودنِ RAG).
+    usedChunks = UsedChunkSerializer(many=True, required=False, default=list)
+    # فقط وقتی DEBUG=True: از کدام collection/فصل چانک‌ها برداشته شد (بازبینیِ داخلی).
+    retrievalDebug = serializers.DictField(required=False)
 
 
 class MessageSerializer(serializers.Serializer):